Why can I not compare oral and injectable milligram figures at all, even roughly?

I keep seeing people reason like this: oral semaglutide 14 mg daily is 98 mg a week, injectable is 2.4 mg a week, therefore the oral dose is roughly 40 times higher, therefore bioavailability must be about 2.5%. That arithmetic is at least internally consistent, and it lands near the published bioavailability figure, so it looks like it works.

But I have also seen it stated flatly that oral and injectable doses cannot be compared even approximately, which contradicts the fact that the naive calculation seems to land in the right region. I would like to know whether the calculation is a valid rough heuristic that happens to have caveats, or whether it is invalid for reasons that make its apparent accuracy a coincidence.

The specific thing I cannot resolve: even if total weekly exposure matched exactly, one route delivers it as a single slow-release depot and the other as seven daily pulses of a variable absorbed fraction. I do not know whether that shape difference matters for a drug whose half-life is about a week, or whether the long half-life makes shape irrelevant.

Worth tabulating the four formulations that come up in this discussion, since the confusion in the question is mostly a consequence of them all being called "semaglutide".

Sources across the rows: [1] [2] [3] [4] [5] [6] [7].

Read down the dose column and you can see why milligram comparisons are hopeless: the same molecule appears at 0.5 mg and at 50 mg, a hundred-fold span, with the low end injected weekly and the high end swallowed daily. The only quantities that travel between rows are effects, and even those need matching populations.

The other thing the table makes obvious: outcome evidence attaches to specific rows, not to the molecule. A renal composite was tested at 1.0 mg injected. A cardiovascular composite was tested at 2.4 mg injected and separately at 14 mg oral. Nobody has tested cardiovascular outcomes at 50 mg oral. Citing SELECT to support a claim about the 50 mg tablet is a formulation-and-dose extrapolation, and it should be labelled as one.

An earlier answer's warning bears repeating in this context: none of these rows describes research-use-only material of unverified content, for which neither the dose nor the delivered exposure is known.

Minor addition on missed doses, since the accumulation arithmetic above has a direct consequence that people get backwards.

With about 9% clearance per day on a daily oral schedule, one missed dose lowers steady-state concentration by roughly the size of one day's input relative to a large accumulated burden - a small perturbation. With weekly injection, one missed dose means about half a half-life passes with no input, which is a much larger relative perturbation, though the long half-life still limits the damage.

Neither of these is a licence to be casual, and both are label questions rather than arithmetic questions - the approved products have specific instructions for missed doses that depend on how much time has passed, and those instructions exist partly to avoid re-triggering gastrointestinal effects by effectively restarting at a dose the person is no longer tolerised to. That second consideration is behavioural and receptor-level, not kinetic, and the arithmetic above says nothing about it.

The asymmetry that matters more is between one missed dose and a run of them. On the daily oral schedule, a week of missed doses removes roughly 1 - 0.906^7 of the accumulated burden, which is about 50% - one whole half-life's worth. So the tablet is forgiving of a single lapse and not forgiving of a lapsed week, and the transition between those two states is fast. On the weekly injectable, the equivalent boundary sits between one and two missed doses.

There is a second-order point buried in this that applies to the oral formulation only. A "missed dose" for a tablet with a fragile absorption mechanism is not a binary event. Taking it with a large glass of water, or ten minutes before breakfast, is a partially missed dose - the tablet was swallowed, the record says it was taken, and the absorbed fraction was a fraction of intended. Adherence measured by pill count therefore overstates exposure adherence for this product in a way it does not for an injectable, and this is one reason trial-grade results and routine-use results diverge more for the oral route.

Trial protocols manage this with explicit dosing instruction and reinforcement at every visit. That is a real component of what produced the PIONEER and SOUL numbers, and it is not something the pharmacokinetics can compensate for.